HiFIVE: High-Fidelity Vector-Tile Reduction for Interactive Map Exploration
Published in ACM SIGSPATIAL 2026 — Riverside, CA (to appear), 2026
Authors: Tarlan Bahadori, Ahmed Eldawy
Accepted at ACM SIGSPATIAL 2026 (Riverside, CA) — camera-ready version forthcoming. A preprint is available on arXiv.
HiFIVE is a data-management framework for scalable, high-fidelity client-side geospatial visualization. It formalizes the visualization-aware tile reduction problem—the trade-off between tile size and visualization distortion—proves it NP-hard, and introduces a practical two-stage solution that keeps interactive maps responsive at terabyte scale.
Why HiFIVE
Many tools for large spatial data fall back on server-side rendering, shipping small images to the client. Users, however, prefer client-side rendering, which allows quick restyling of the data for a better exploration experience—but that requires sending the data itself, and full-fidelity vector tiles are far too large. HiFIVE closes this gap by reducing tiles in a way that is aware of how they will ultimately be drawn.
Key Capabilities
- Formal problem definition of visualization-aware tile reduction, with an NP-hardness proof
- Two-stage reduction: triage followed by sparsification
- Selective pruning of records, attributes, and values using information-theoretic and spatial criteria
- Substantial tile-size reductions while preserving visual fidelity and interactive performance at terabyte scale
Artifacts:
- 📄 Preprint: arxiv.org/abs/2603.10270